CF analysis

1 Objectives

• Using well data in the 10 miles buffer, do CF analysis with:
  • Panel data
    • For LR data, owner-level
    • For TB data, well-level
  • Cross-sectional data
    • For LR data, owner-level
    • For TB data, well-level

2 Panel data

2.1 Data in the 10 miles buffer

reg_data_10mi_raw <- 
    here("Shared/Data/WaterAnalysis/reg_data_10mi_raw.rds") %>%
    readRDS()

# - Data on TB - #
TB_10mi <- reg_data_10mi_raw[nrdname=="Tri-Basin",]
# - Data on LR - #
LR_10mi <- reg_data_10mi_raw[nrdname=="Lower Republican",]

2.1.1 LR Data

• The data is aggregated by nrd_owner_name and year.
• In aggregating the data, the variable should be taken by mean weighted by acres.

LR_10mi_By_y_owner <-
  LR_10mi %>%
  .[,.(
    usage = weighted.mean(usage, acres/sum(acres)),
    treat2 = mean(treat2),
    # --- soil --- #
    silttotal_r = weighted.mean(silttotal_r, acres/sum(acres)),
    claytotal_r = weighted.mean(claytotal_r, acres/sum(acres)),
    slope_r = weighted.mean(slope_r, acres/sum(acres)),
    ksat_r = weighted.mean(ksat_r, acres/sum(acres)),
    awc_r = weighted.mean(awc_r, acres/sum(acres)),
    # --- weather --- #    
    pr_in = weighted.mean(pr_in, acres/sum(acres)),
    pet_in = weighted.mean(pet_in, acres/sum(acres)),
    gdd_in = weighted.mean(gdd_in, acres/sum(acres)),
    # --- tr --- #
    cntyname_fix = unique(cntyname_fix)
  ), by = .(nrd_owner_name, year)] %>%
  .[,county_year := paste0(cntyname_fix, "_", year)]

2.1.2 TB Data

• For TB data, keep using yearly well-level data.

se_vars <- names(LR_10mi_By_y_owner)

TB_10mi_By_y_well <-
  TB_10mi %>%
  .[,county_year := paste0(cntyname, "_", year)] %>%
  .[, ..se_vars]

2.1.3 Regression Data

• Merge LR data and TB data.

reg_data_10mi <- bind_rows(LR_10mi_By_y_owner, TB_10mi_By_y_well)

2.2 CF Analysis

2.2.1 1st CF

2.2.2 2nd CF

3 Cross-sectional data

• The cross-sectional unit is:
  • For LR data, owner
  • For TB data, well-id
• About tr=="5_22"
  • In these areas in TB, 9 acre-inches regulation has started from 2009 (the second year of the allocation period 2008-2012)
    • two options for those data points
      • remove entire data points related to tr=="5_22"
      • use only the data points starting from 2009

3.1 Aggragate annual data to whole-periods data by taking mean

# === LR data === #
LR_10mi_By_owner <- 
  LR_10mi_By_y_owner %>%
  .[,.(
  usage = mean(usage),
  treat2 = mean(treat2),
  # --- soil --- #
  silttotal_r = mean(silttotal_r),
  claytotal_r = mean(claytotal_r),
  slope_r = mean(slope_r),
  ksat_r = mean(ksat_r),
  awc_r = mean(awc_r),
  # --- weather --- #  
  pr_in = mean(pr_in),
  pet_in = mean(pet_in),
  gdd_in = mean(gdd_in),
  # --- tr --- #
  cntyname = unique(cntyname_fix)
  ), by = .(nrd_owner_name)] %>%
  .[,nrd_owner_name := NULL]


# === TB data === #
TB_10mi_By_well <-
  TB_10mi %>%
  # remove data points related to tr=="5_22"
  .[tr!="5_22",] %>%
  .[,.(
  usage = mean(sum(volaf*12)/sum(acres)),
  treat2 = mean(treat2),
  # --- soil --- #
  silttotal_r = mean(silttotal_r),
  claytotal_r = mean(claytotal_r),
  slope_r = mean(slope_r),
  ksat_r = mean(ksat_r),
  awc_r = mean(awc_r),
  # --- weather --- #  
  pr_in = mean(pr_in),
  pet_in = mean(pet_in),
  gdd_in = mean(gdd_in),
  # --- tr --- #
  cntyname = unique(cntyname)
  ), by = .(wellid)] %>%
  .[,wellid := NULL]

# unique(TB_10mi_By_well$treat2)

agg_reg_data_10mi <- rbind(LR_10mi_By_owner, TB_10mi_By_well)

3.2 CF Analysis

3.2.1 1st CF

3.2.2 2nd CF